An optimal dynamic KC<SUB align="right">i-slice model for privacy preserving data publishing of multiple sensitive attributes adopting various sensitivity thresholds

Abstract

Optimal KCi-slice model is an extension of KC-slice, KCi-slice and Novel KCi-slice models for publishing the data with multiple sensitive attributes. The proposed Optimal KCi-slice model imposes the privacy threshold only on high sensitive values instead of applying on all the sensitive values of each sensitive attribute. It imposes the necessary privacy threshold to each sensitive attribute based on their sensitiveness. It automatically leads to a good utility rate and required privacy levels on all high sensitive values of each sensitive attribute. Optimal KCi-slice model finishes the data publishing process in two steps. Firstly, it uses the Enhanced semantic l-diversity algorithm to attach the tuples into the buckets and splits the sensitive attributes into several sensitive tables. The second step decides the correlated quasi attributes and also concatenates the correlated quasi attributes with SIDs (Sensitive bucket identifiers) of sensitive buckets. Proposed Optimal KCi-slice model achieves a high utility rate and required privacy levels compared to all the existing models.